How to choose a B2B web design agency

A website is one of the few purchases where you cannot inspect the product before you buy, cannot easily judge quality after you receive it, and will not know whether it worked for another six to twelve months. That combination is unusual, and it is why the web design market contains such a wide spread of outcomes for a similar price.

This is written from the supply side — we are a studio, and we lose pitches to agencies we think are worse than us and win some we probably should not have. What follows is the set of signals we would use if we were buying instead of selling.

First, decide what kind of problem you have

The single most expensive mistake is hiring the right agency for the wrong problem. Three very different problems get described with the same sentence — "we need a new website":

A positioning problem

Your website is clear, fast and tidy, but it says something unconvincing. Prospects arrive and do not understand why you rather than the other four on the shortlist. A redesign will not fix this. What you need is a positioning and messaging engagement, which may or may not end in a new site.

A presentation problem

You know exactly what you are and who for, and the website undersells it. It looks like a 2019 template, the case studies are thin, the site is slow on a phone. This is the problem a good design and development studio solves, and it is the most common one.

A systems problem

The website is fine as a brochure but cannot do what the business now needs — a portal, a configurator, a gated resource library, a proper integration with your CRM. This is a product build wearing a website's clothes, and it should be scoped and priced as one.

Write down which of the three you have before you speak to anyone. If an agency's first response does not acknowledge the distinction, they are going to sell you the thing they build most often.

How to read a portfolio properly

Portfolios are curated, obviously. What is less obvious is that the best ones tell you very little about whether a project will go well, because a beautiful screenshot is evidence about the designer and almost nothing about the process, the project management, or the state of the codebase underneath.

Four things make a portfolio actually informative:

1. Visit the live sites, on your phone, on your own network. Roughly half the beautiful sites in agency portfolios are unpleasant to use on a mid-range Android on a train. That gap is the whole story about the agency's engineering standards.
2. Check whether the sites are still live and still theirs. If three of the five flagship projects have been rebuilt by someone else since, ask about it. Sometimes the answer is fine — the client was acquired, the brand changed. Sometimes it is not.
3. Look for a project like yours in shape, not in sector. A complex multi-market site with a real CMS and 200 pages is a fundamentally different engagement from a gorgeous eight-page site for a design studio. Sector similarity matters far less than structural similarity.
4. Read the case studies for decisions, not features. "We used GSAP for scroll animations" tells you what they did. "The client's buyers were comparing four vendors on a spreadsheet, so we put the comparison on the page ourselves" tells you how they think.

And one anti-signal: a portfolio where every site looks like the same site with different logos. Consistency of craft is good. Consistency of layout means you are buying a template with a bespoke price tag.

The questions that actually separate agencies

Most "questions to ask your web agency" lists are things any agency can answer smoothly. These are harder to fake because the answer either exposes a real process or reveals there is not one.

"Who exactly will do the work, and what else are they on that month?"

You are looking for named people and an honest answer about capacity. The bad answer is a vague "our team". The worst outcome in this industry is being sold by seniors and delivered by whoever was free.

"What happens when we disagree about a design decision?"

A studio with a real process has an answer involving evidence and a decision-maker. A studio without one says "we'll keep iterating until you're happy", which sounds accommodating and is how projects lose four months.

"Show me a project that went badly and what you changed afterwards."

Everyone has one. An agency that cannot name one is either very new or not being straight with you. The content of the answer matters less than whether it exists.

"What will you need from us, and by when?"

Web projects are late because of content and approvals far more often than because of design or code. A studio that has run real projects will hand you a list and a schedule unprompted. One that says "just the basics" has not thought about it.

"Who will own the code, the accounts and the domain?"

The only acceptable answer is you. If the site lives in the agency's Webflow workspace, their hosting account, their repository, or a theme licensed in their name, you are renting your own website.

"What does month one after launch look like?"

Launch is the middle of the project, not the end. Ask about the redirect map, the Search Console handover, the analytics check, the fixing of whatever the first fortnight reveals. If this is all billed as a surprise extra, factor that in.

What the money actually buys

Price variance in this market is enormous and mostly explained by three things: how much senior time goes into the project, how much of it is strategy versus production, and how much risk the agency is absorbing on your behalf.

There is no correct band. There is only a mismatch between band and problem, which is what to avoid. A $40,000 process applied to an eight-page site produces an expensive eight-page site.

See our published prices — We publish ours, including what is and is not included, because a price behind a form wastes everybody's afternoon.

The specialist question: agency, studio, or freelancer

These are genuinely different purchases and each is right sometimes.

A large agency

Buy this when the project spans brand, campaign, media and web, or when your organisation needs the reassurance and process of a big supplier. You pay for coordination, and coordination is a real thing you are sometimes buying.

A small studio

Buy this when the work is the deliverable and you want the people who pitched to be the people building. Typically the best ratio of senior time to money. The risk is capacity — a studio of four has no bench.

A freelancer

Buy this for a defined, contained piece of work where you are effectively the project manager. Excellent value, and completely dependent on one person's month not going wrong.

A tell that cuts across all three: who answers your technical questions? If the person in front of you can explain why they would choose one platform over another for your specific case, without reaching for a partner badge, you are talking to someone who builds things.

Contract terms worth insisting on

None of these are unusual. An agency that resists all of them is telling you something.

• Intellectual property in the design and the custom code transfers to you on final payment, in writing.
• All accounts — domain, hosting, CMS, analytics, Search Console — are created in your name, or transferred to it at launch.
• A defined number of revision rounds at each stage, with what counts as a round written down. Unlimited revisions is not generosity; it is an absence of process.
• A named scope document, and a stated way to change it that involves a number and a date rather than an argument.
• A payment schedule tied to milestones rather than to dates, so a project that stalls on their side does not invoice on schedule.
• A warranty period after launch — 30 days is common — where things that are broken get fixed without a new invoice.

Red flags that are worth walking away over

• Guaranteed SEO rankings. Nobody can guarantee a position on a platform they do not control, and the people who claim to are describing something else — usually paid ads, or a ranking for a phrase nobody searches.
• A price quoted before anyone has asked what the site needs to do. It means there is a template and the price is for the template.
• Pressure to sign this month for a discount. Studios with a full pipeline do not do this.
• No questions about your customers. An agency that asks only about your taste is going to design something you like and your buyers do not.
• Hosting that can only be theirs, and a refusal to explain why.
• A proposal with no dates in it, or with a single date at the end.

How to de-risk the whole thing

The most effective mechanism is small and boring: buy something little first. A single landing page, a design concept for the homepage, a performance audit of your current site. You will learn more from one small paid engagement than from three rounds of pitching, because you see how they actually work — how they ask questions, how they handle feedback, whether the thing arrives when they said it would.

It costs a few hundred to a couple of thousand, and it tells you what a reference call cannot. We offer it for exactly this reason, and so do most studios worth hiring. If an agency will not sell you a small piece of work first, ask yourself why they need the whole project committed up front.

A short version

1. Name your problem: positioning, presentation, or systems. Do not let the agency name it for you.
2. Judge portfolios on live phones, for structural similarity, and on how the case studies explain decisions.
3. Ask who does the work, what happens when you disagree, and what they need from you and when.
4. Match the price band to the problem, and find out how much of the fee becomes senior making time.
5. Insist on owning the IP and every account.
6. Buy something small first.